Amp Burn-In: What actually happens? - Page 3

I've done experiments similar to this. When I got my Gilmore Lite I compared it frequently against my Gilmore V2 (old gear, toasty burnt in). The difference in bass response was very noticeable until finally, one day I plugged it in and bingo, they were virtually identical for bass response. In a simlilar vein, I've been listening back and forth between my GS-1 and my Gilmore V2-SE as the GS-1 was burning in and again it's been very noticeable how the sound has been changing and they've been getting more and more similar with time. Rigorous enough for a scientist -> Heck no! Rigorous enough for what I'm willing to do? -> Yes!

Based on what? The assumption that a builder uses, for example, Blackgate caps in a system because they have tested every other kind of cap and found them to be the best sounding. Are you a amp builder that has put together two identical circuits, one with BlackGate, one with Rat Shack caps and tested them to see if you can hear a difference? Can you refer me to somewhere where someone else has done this? Or a manufacture that gives that as a reason. I would be very interested to see that.
I have no doubt that BlackGates are a very good cap, and I might as a engineer spec them in a component where I wanted low failure, and very consistent QC. In a audio circuit I would bet they are put there because people recognize the name and are more likely to pay what you want.
--
Dana

Thank you for the thread. Very interesting. I will see if I can hear this on my 650s.

This is a story about headphone amp burn-in:

http://www4.head-fi.org/forums/showthread.php?t=24893

It's not the last time I've encountered changing characteristics with electronics -- sources and amps. I'm a firm believer in break-in effects. Like others, I always use a sonic reference to make sure that it's not just a mental thing. Of course that's still no guarantee. But after all electronics developers seem to agree on the issue, generally speaking. Unfortunately I have nothing to contribute when it comes to explanations

Yes, for instance I have had the Gilmore V2-SE with both Panasonic and Black Gate Capacitors in it. I have also compared my Gilmore Balanced Reference (with Black Gates) against another user's Gilmore Balanced Reference without Black Gates. Any other questions? Quid Pro Quo -> Have you tried building a circuit with all capacitors available and listened to them to determine they sound the same? If not, how can you make these statements? I could see you advancing a hypothesis, but not having a conclusion formed before you even tried which is the manner in which I am interpreting your posts. Please forgive me if this is not the correct meaning that I should have taken from your posts. From my side, I only have to hear two sound different for my earlier statements to be correct while you would need to listen to all of them to prove that they do not sound different.

And were you blind to this difference otherwise? That is, did you know which one you were listening to? If you were not blind to the difference, then the difference is at least potentially a placebo effect.

Saying "I've listened and I've heard the differences" is the common, popular argument (not just from you, gpalmer, I don't mean to pick on you specifically). It is unfortunately not proof of a real difference, unless the listener is blind to what the change is.

Placebo effects are enormously powerful. People experience real, physical changes from placebo drugs--improvements from disease, side effects, the whole deal. In fact, people who are susceptible to placebo effects with pain killers actually change the composition of neurotransmitters in their brains (effectively releasing endorphins), which really does stem the pain. It isn't just some "all in your head" nonsense experienced only by the weak-minded. These kinds of things are why placebo controlled-studies are mandatory before the FDA will approve a drug.

Does anyone really think that if your brain can measurably change biochemistry due to expectation effects that your perceptions of sound reproduction are immune from them? Seriously?

Yes, it's possible in many cases that there is a real difference and there really is break-in of some kind--I don't dispute that. And, in many cases, there are perfectly well-understood reasons for such effects. But in the absence of such an explanation, saying "just because we don't know the reason doesn't mean it doesn't exist" is a weak, weak argument (even if it sometimes turns out later to be true). We do know placebo effects exist and they're powerful.

For some of us to believe it's not a placebo effect, we'd like a reasonable explanation of what the underlying cause of the break-in effect is, and it has to be something better than "I've heard it so it must be real." I have yet to see one here.

Good luck. You'd be amazed at how resistant the audio world is to basic scientific methodology. It's not hard to find grand treatises on why blind testing is flawed and wrong, none of which would pass muster in even vaguely rigorous scientific journals. Noticed the "DBT-free" tag on the cables forum? Essentially, this means that "we believe this without scientific evidence and our minds are made up about it--so don't confuse us with facts." OK, that's probably too harsh--people probably just don't want the threads constantly torpedoed by the same questions over and over because that gets pretty old. But it says a lot that the stance taken is banishment rather than honest discourse.

Buy 3 identical SR71ss.
Put one in a Faraday cage with controlled temperature conditions.
Velcro the other two together, toss the bundle into your rucksack and only use ONE of them for a month.
Then give all three of them all to me

Yeah, Black Gates sound just like Rat Shack capacitors . . . and my alarm clock CD player sounds just like a Meridian G08 . . . and chuck steak tastes like filet mignon . . .

Yes, placebo can be a real issue, but almost everyone who spouts the placebo mantra has never listened to anything. I'm not necessarily saying you fall into this category (although others on this thread do, according to their own admission), and I don't want to argue the problems with the placebo argument again, but, in brief, it can't explain everything that people claim to hear.

I don't see how that's relevant. Blind testing is the only sure way to control for expectancy effects, whether you've "listened to anything" or not. Attacking the questioner is still a logical fallacy--you have to address the argument made, not who made it.

Now, I'm certainly willing to grant that experienced listeners know better what to listen for and can almost certainly make discriminations that others can't--I am completely willing to believe that JaZz can detect differences in amps which I cannot (which is almost certainly good for my wallet ). If I were to run a blind test on amp break-in, I'd certainly choose people like him over people like me because there'd surely be a better chance of finding an effect. I don't dispute that.

However, that does not mean that experts are immune to expectancy effects. 30+-year experts at wine tasting have failed blind tests for wines which they claimed beforehand were "obviously" different.

So, why should effects based on questionable physics be given the benefit of the doubt when a perfectly reasonable alternate explanation exists?

Let me get this straight--the belief that one has received a painkiller can produce a measurable biochemical change in one's brain (and a much larger change in perceived pain experienced), but the expectation that two things will sound different cannot explain differences in sound perception? I'll admit I'm skeptical, but I'm eager to hear alternatives. Could you provide an example or two?

You've hit on two key points that need emphasis.

First, in the case of the wine experts, the wines were different in reality. The experts simply didn't detect the differences. A blind test is an excellent way to determine that perceived differences are real. It is a lousy way to determine that differents that are not perceived are not real. That is, failure to detect differences on a blind test does not imply that any differences are not there. A blind test is only interpretable in the case of a positive result.

Second, of course expectancy plays a role in perception, and probably a bigger one than you realize. Perception is not a veridical translation of stimulation of some receptor to a locus in the brain. Downflow from the brain can alter the input, so that perception is physically altered before it even reaches the brain for processing. However, the "expectancy" effect needs to be systematic in some way. For an expectancy effect to be present, an expectancy must be present (this may seem obvious, but think about it for a bit). If the result of the test is different than the a priory expectancy, then the "expectancy" effect cannot be used as an explanation of the result, at least with any credibility.

Come to think of it, this actually creates a more sensitive test for expectancy effects than blind testing: the use of a false expectancy. If expectancy is producing the effect, the one that the person expects to be better should consistently be judged as better. So, creating a false expectancy ("Cable A costs 20 times as much as Cable B", when the opposite is true) should create a preference for Cable A. If this does not occur, or if listeners appear to prefer cable B anyway in a properly designed study, you can kiss expectancy effects goodbye. Of course, this cannot be done in a "blind" experiment, as the whole idea is to study the effects of expectancy in an unblinded situation. Note that this is not an attempt to control for expectancy effects. It's more of an effort to study them directly and determine if their effect on audio judgements is worth bothering with in the first place. Why bother to control for an expectancy effect if you can't deliberately produce one to demonstrate its existence?

Here's a couple of answers/issues. (1) How does the placebo effect cause a died-in-the-wool skeptic who is absolutely convinced that a certain difference does not exist to become a believer when he hears the difference? This is reported over and over. We repeatedly hear people who are convinced when they listen, even people who were sure there were no differences. I mean, the people who generaly argue placebo are NOT the people who have listened and found no difference (with a very few exceptions). It is always claimed by the people who have never listened (and refuse to listen), but who claim there can't be a true audible difference? Doesn't this tell us something? And, with the respect to the experiences of skeptics turned believers, is the placebo effect so powerful that no commited skeptic on the planet earth can possibly resist it?

(2) How does the placebo effect cause one to prefer the cheaper cable or component (i.e., the result of the listening experiment is exactly opposite the "expectation")? I have experienced this, and others on other threads have reported this phenomenon also.

(3) What's the explanation for the reported phenomenon of people reporting differences between cables, for example, when they did not realize a cable had been changed? I have also experienced this and others have also. In this case, there would seem to be no "expectation" at all, and yet a difference is perceived.

As for blind testing, this has also been discussed before, but I think there are problems with the typical blind test, which don't really mirror real world listening conditions. Consequently, you get the John Atkinson result, where he listens to amps in a blind test, can't pick out the cheap one from the expensive one, buys the cheap one becuase he is convinced that it sounds the same as the other amps, and then is extermely disapointed with the sound of his system for the next year. I think I would probably fail blind tests, for example, trying to distinguish different CDP's. But let me live with one for a month, then change it without my knowledge, and I'll be able to tell the difference if you give me some time to listen to it. But I won't be able to tell the difference flipping back and forth every 30 seconds with music I don't listen to and with the CDP's connected to other components that are not part of my system.

Anyway, my .02.

That's only sort of true. If you specify the minimum acceptable effect size, you can compute the probability that the true population difference is actually smaller than that, which can support the conclusion that there's no (meaningful) effect. This is pretty technical and requires somewhat large sample sizes to accomplish meaningfully, but it's not impossible.

A bigger one than I realize? Probably not.

Depends. If the expectancy is simply "something is different" (which is sometimes the case in medical studies and experimental psych), you still get (false) perceptions of difference. We get this in the lab all the time.

This is kind of a fun idea, actually, pitting perception against expectation. Of course, if you get a general preference for Cable A, then you have evidence that expectations clearly are driving preference. It would actually be even more sensitive if you ran multiple conditions: one group of subjects is told (incorrectly) that A is more expensive, one that isn't told anything, and one that is told B is more expensive. Comparing results would give you an estimate of the size of the expectancy effect.

Perception (and memory) are not constant, they are affected by all kinds of factors like fatigue, stress, habituation, noise, etc. People may perceive some variation, then interpret that variation as being due to something that it isn't. Incidentally, this routinely happens in medical studies. I have a friend who's a CDC epidemiologist and she's read all kinds of fascinating comments from people in placebo conditions. Stuff like "I didn't think the drug would do anything because every other drug I've taken for this condition didn't, but this one really helped, except for side effects XYZ" etc. In our lab we get wild descriptions/explanations of "patterns" in stimuli which were randomly generated (and we tell subjects this).

I never claimed that all differences are placebo. (In fact, I never even claimed that amp burn-in differences are--just that, up to now, I haven't heard a better explanation.) Some of these may be real differences. I find your (3) case much more compelling than random "I swear I've heard it" reports which are much more common, though care has to be taken to ensure nothing else besides the stated change has occurred. The "I didn't know it was going on" seems pretty unlikely in the case of burn-in, though.

Incidentally, who's going around and messing with your cables without telling you? I hate it when other people fiddle with my rig...

Also incidentally, this kind of thing happens to me all the time with my speaker rig--I find something sounds wrong, and then I'll realize that somebody (usually me) has bumped one of the speakers so it's moved a little. (I don't usually notice small displacements, but small rotations.)

I don't disagree with this at all--a totally legit complaint. That's a methodological problem which is not intrinsic to blind testing, however, but to particular implementations of it. It's cheaper and easier to do a blind test in a truncated way, but it isn't necessarily better.

Fun discussion!

This is true, unless your talking about speakers and headphones.